Commentary: The Limits of Risk Factors Revisited Is It Time for a Causal Architecture Approach?
Open accessThe DOI matches its Crossref record; 10 of the 10 references it cites were found in a registry or an index.
Abstract
NEW CHALLENGES FOR THE 21ST CENTURY Epidemiology today stands at a juncture. Our methods are increasingly sophisticated, and ever-greater pools of data are being collected and could be available for epidemiologic inquiry. Yet the core approaches inherent to, and the utility of, epidemiology are being questioned perhaps as never before by a growing number of stakeholders.1–3 Big data, precision medicine, various -omics initiatives, technology-informed intervention, and other new research horizons leave open questions about where epidemiology does (and does not) fit in. The methodologic advances in our field in the past quarter century have led to the establishment of an armamentarium that allows epidemiology to take its place alongside other sciences with their own well-established, methodologic conventions. They also have led to a particular orientation we take in our work. Our textbooks and advanced epidemiology coursework predominantly instruct our students on how to estimate risk factors, and many of our methodological tools for confounder control provide us with increasingly sophisticated ways to estimate these risk factors more precisely in the presence of complex data structures and potential sources of confounding. There is little doubt that traditional risk factor epidemiologic methods have contributed in critical ways to our understanding of the effects of smoking, environmental toxins, pre- and peri-natal exposures, and viral agents of infectious and chronic disease, among others. However, it is also clear that many other public health successes of the past century owe little to our methodological developments as a discipline, and that there is growing concern that an effort to increasingly identify small effects marginalizes us as a field.3–5 WHY WE CANNOT (SERIOUSLY THIS TIME) CONTINUE RISK FACTOR APPROACHES The calls for a move away from risk factors have been compelling and numerous over the past 20 years. It is time to heed the growing drumbeat of concern about these risk factor approaches and propose an alternate approach to how we may conceptualize our hunt for causes of disease, because, in part: (1) these approaches lead us to focus on even more precise estimates of risk factors that we already know are important, or alternatively, already know are not very important; (2) these approaches encourage us to continue to pursue a biomedical model of disease; and (3) the world is too complicated to distill into a set of correlates. These are not new ideas, but perhaps worth restating, with citations from some seminal papers that articulated some of these key ideas over the past decades. First, the traditional risk factor approach keeps us bound to what Beverly Rockhill termed a decade ago as “the increasingly reductionist hunt for causes.”4 Because the field of risk factors is relatively well ploughed and we have a set of “usual suspects” (poverty, smoking, diet, toxins), we are often left estimating risk ratios of relatively small magnitude and trying to make the case for their role in public health. One needs only to look toward the field of genetic epidemiology to see this reductionist hunt in action. Furthermore, we attempt to refine and reframe the extent to which previously identified associations are causal with increasingly restrictive samples with precise analyses. As just one common example, we apply propensity score matches to the data and only analyze those pairs or sets for which there is a match. This is entirely reasonable if the goal is to isolate the causal effect of an exposure on an outcome; only those pairs with a match on the propensity to be exposed contribute to our ability to assess the causal impact of the exposure. But if our overall goal is to improve population health, propensity score matching is perhaps much less useful if those pairs who match on propensity to be exposed are not representative of the population in which we would like to intervene, and if the factors that create extremes in the data and for which no match is possible are also factors that interact with exposures of interest. Second, we often, and perhaps increasingly, remain “prisoners of the proximate,”6 as Anthony McMichael termed, and we are almost 20 years ago away from when Shy7 provocatively implored us to shift from biomedical models of disease to models that incorporate the complex social world through which disease generates. We are further still away in time from other scholars who have encouraged us to consider the data-generating mechanisms through which exposures become embedded if we want any hope in changing population health.8–10 Have we effectively heeded the calls for a movement away from proximate risk factors? In some ways, yes. Rapid progress in areas such as social epidemiology and lifecourse epidemiology have challenged the dominant risk factor paradigm as potentially inadequate to the questions that are of central interest to current public health problems. In the ensuing ye
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Sick individuals and sick populations
Rose, G. (2001) · International Journal of Epidemiology · doi:10.1093/ije/30.3.427 · Cited by 3,736
Sick individuals and sick populations
ROSE, G. (1985) · International Journal of Epidemiology · doi:10.1093/ije/14.1.32 · Cited by 2,701
Pepe, M. S. (2004) · American Journal of Epidemiology · doi:10.1093/aje/kwh101 · Cited by 1,276
Why Representativeness Should Be Avoided
Rothman, K. J., Gallacher, J. E., & Hatch, E. E. (2013) · International Journal of Epidemiology · doi:10.1093/ije/dys223 · Cited by 861
Hood, L., & Flores, M. (2012) · New Biotechnology · doi:10.1016/j.nbt.2012.03.004 · Cited by 648
Prisoners of the Proximate: Loosening the Constraints on Epidemiology in an Age of Change
McMichael, A. J. (1999) · American Journal of Epidemiology · doi:10.1093/oxfordjournals.aje.a009732 · Cited by 502
Compound Treatments and Transportability of Causal Inference
Hernán, M. A., & VanderWeele, T. J. (2011) · Epidemiology · doi:10.1097/ede.0b013e3182109296 · Cited by 297
What is a representative brain? Neuroscience meets population science
Falk, E. B., Hyde, L. W., Mitchell, C., Faul, J., Gonzalez, R., Heitzeg, M. M., Keating, D. P., Langa, K. M., Martz, M. E., Maslowsky, J., et al. (2013) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1310134110 · Cited by 265
FTO gene polymorphisms and obesity risk: a meta-analysis
Peng, S., Zhu, Y., Xu, F., Ren, X., Li, X., & Lai, M. (2011) · BMC Medicine · doi:10.1186/1741-7015-9-71 · Cited by 210
Formalizing the role of agent-based modeling in causal inference and epidemiology
Marshall, B. D. L., & Galea, S. (2015) · American Journal of Epidemiology · doi:10.1093/aje/kwu274 · Cited by 192
Self-citation
Cited by
The social determinants of mental health and disorder: evidence, prevention and recommendations
Kirkbride, J. B., Anglin, D. M., Colman, I., Dykxhoorn, J., Jones, P. B., Patalay, P., Pitman, A., Soneson, E., Steare, T., Wright, T., et al. (2024) · World Psychiatry · doi:10.1002/wps.21160 · Cited by 1,253
THIMM‐KAISER, M., BENZEKRI, A., & GUILAMO‐RAMOS, V. (2023) · The Milbank Quarterly · doi:10.1111/1468-0009.12642 · Cited by 114
Colder Carras, M., & Kardefelt-Winther, D. (2018) · European Child & Adolescent Psychiatry · doi:10.1007/s00787-018-1108-1 · Cited by 99
Zanini, M. J., Domínguez, C., Fernández-Oliva, T., Sánchez, O., Toda, M. T., Foraster, M., Dadvand, P., & Llurba, E. (2020) · Current Hypertension Reports · doi:10.1007/s11906-020-01088-4 · Cited by 29
Metadata
- Authors
- Keyes, K. M., & Galea, S.
- Venue
- Epidemiology
- Published
- 2017-01
- Volume
- 28
- Issue
- 1
- Pages
- 1-5
- ISSN
- 1044-3983 (ISSN-L)
- OpenAlex
- W2531631555
- Cited by
- 62
- References
- 10
- FWCI
- 26.68
- Type
- journal article
- Language
- en
- Open access
- green
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Commentary: The Limits of Risk Factors Revisited Is It Time for a Causal Architecture Approach? https://app.cochat.ai/papers/10.1097/ede.0000000000000578
- The paper
- Keyes, K. M., & Galea, S. (2017). Commentary: The Limits of Risk Factors Revisited Is It Time for a Causal Architecture Approach? Epidemiology. https://doi.org/10.1097/ede.0000000000000578
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.